Google unveils Gemini 3.5 Transcribe, a faster and cleaner AI speech-to-text model

Google has introduced Gemini 3.5 Transcribe, an AI model designed to turn voice input into polished text by removing filler words and corrections. The model already powers the Gboard Rambler feature on the Pixel 11 and will soon roll out across Google's ecosystem.

Google says it is about 70 percent faster than its previous Chirp 3 engine, with the live-speech error rate dropping to 5.5 percent from 7.32 percent. While the accuracy gain is modest, the speed improvement makes voice typing less tedious.